Anthropic releases Claude Fable 5.2 for all paid users and restricted Mythos 5.2

In short

• Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on 9 November 2026.
• The public version is available now on paid plans at $8 per million input tokens and $40 per million output tokens, down from $10 and $50.
• Mythos 5.2 shares the same model but is limited to organisations approved for cybersecurity or biomedical research.
• Anthropic says the models are up to 30% faster, with benchmark gains in coding, reasoning and vision from the company's own tests.

Anthropic launched Claude Fable 5.2, plus Claude Mythos 5.2, two new frontier artificial intelligence models, on 9 November 2026. Claude's official X account said the model is now offered across all paid plans, the Claude Platform and the main cloud services, with input priced at $8 per million tokens and output at $40 per million, reduced from $10 and $50; the thread links to Anthropic's announcement. Mythos 5.2, by contrast, is restricted to vetted organisations.

How do the two new models differ?

What Anthropic says the models can do

How developers reacted to the launch

AI Statute (@StatuendState) called the launch "a strong incremental release", saying on X that Fable 5.2 and Mythos 5.2 look like a step up on hard, long-horizon technical work rather than a general-purpose leap.

深夜提灯人 (@bljping1) asked on X in Chinese why Anthropic recommends the previous generation.

Keith (@Keith985211) wrote on X in Chinese that he has joined Claude Duet but finds the output speed far too slow.

What does the launch mean for Claude users?

Benchmark results for both models

Claude's new models: key questions

Is Claude Mythos 5.2 available to everyone?

No. Claude Mythos 5.2 is limited to organisations approved through Anthropic's Life Sciences and Cyber Verification Programs. The public version is available now on all paid plans, the Claude Platform and major cloud platforms.

How much does the public version cost?

The public version costs $8 per million input tokens and $40 per million output tokens, down from $10 and $50 for Fable 5.1. The price applies on paid plans, the Claude Platform and major cloud platforms, according to Claude's X thread.

What is the public version designed to do?

Anthropic describes the model as built for ambitious, open-ended technical work that calls for sustained reasoning. On the company's own tests it made gains in coding, reasoning and vision, and it helped formalise the proof of Fermat's Last Theorem in 30 hours, according to Anthropic.

Will average Claude users notice a difference?

For everyday Claude subscribers, the public model is available on paid plans; Mythos 5.2 remains limited to organisations approved by Anthropic.

At InvestIn.News, we think the 20% cut in API pricing matters more to most builders than the benchmark scores. The public model now charges $8 per million input tokens and $40 per million output tokens, down from $10 and $50, which makes frontier-level coding more affordable for small teams and on-chain agent developers.

Splitting one underlying model into a safeguarded public version and a vetted-only Mythos is also a measured way to release risky capabilities: Fable 5.2 keeps strict cybersecurity and biology guardrails, while Mythos 5.2 stays with approved research organisations.

The claimed gains, however, come from Anthropic's own evaluations. Independent tests are not yet available, so we treat the benchmark leap with appropriate caution. We will revisit this view once developers publish real-world coding results and independent labs assess the models.

Follow the latest as it happens on AI NewsFeeds.

What to watch: Mythos 5.2 is not generally available yet; interested organisations can apply through Anthropic's Life Sciences and Cyber Verification Programs.